WO2023228542A1

WO2023228542A1 - Authentication system and authentication method

Info

Publication number: WO2023228542A1
Application number: PCT/JP2023/012047
Authority: WO
Inventors: 鉄平福田; 亮太藤井; 慎太郎岡田
Original assignee: パナソニックＩｐマネジメント株式会社
Priority date: 2022-05-27
Filing date: 2023-03-24
Publication date: 2023-11-30
Also published as: JP2023174185A

Abstract

The authentication system comprises: an acquisition unit that acquires an audio signal of a speaker's spoken voice; a detection unit that detects a first utterance interval spoken by the speaker from the audio signal and a second utterance interval spoken by the speaker from the audio signals of a database in which are registered audio signals of a plurality of speakers; a determination unit that compares a first audio signal of the first utterance interval and a second audio signal of the second utterance interval and determines authentication conditions for authentication using the first audio signal based on the length of the second audio signal of the second utterance interval or the number of sounds included in the second utterance interval; and an authentication unit that authenticates the speaker.

Description

Authentication system and method

The present disclosure relates to an authentication system and an authentication method.

Patent Document 1 discloses a telephone communication device that registers voiceprint data for voiceprint authentication from voice received during a telephone call. The telephone device acquires the received voice, acquires the telephone number of the speaking party, and extracts voiceprint data from the acquired voice. Next, the telephone device measures the acquisition time of the received voice. The telephone device determines whether the total acquisition time length of at least one voiceprint data in the telephone directory and corresponding to the same telephone number as the acquired telephone number is longer than the time required for voiceprint verification. If the telephone device determines that the total acquisition time of the voiceprint data is longer than the time required for voiceprint verification, the telephone device associates the acquired telephone number with the voiceprint data and stores them in the storage unit.

Japanese Patent Application Publication No. 2016-53598

In Patent Document 1, when the total acquisition time length of the speaker's voiceprint data exceeds a predetermined value, the voiceprint data is registered in a database in association with the speaker's telephone number. In other words, the telephone device disclosed in Patent Document 1 always requires voiceprint data for registration used in voiceprint authentication for a total acquisition time longer than a predetermined value. Therefore, the user needs to speak for more than a predetermined time to register the voiceprint data, and the user also needs to speak for the same amount of time for voiceprint authentication, which improves user convenience. Improvements are expected.

The present disclosure has been devised in view of the conventional situation described above, and improves user convenience by determining the speaking time during authentication according to the total length of the user's uttered audio acquired at the time of registration. With the goal.

The present disclosure includes an acquisition unit that acquires an audio signal of a speaker's utterance, a first utterance section in which the speaker is speaking from the acquired audio signal, and audio signals of each of a plurality of speakers. a second speech section in which the speaker is speaking from the speech signal of the database in which is registered; a first speech signal of the first speech section; and a detection section that detects the second speech section of the first speech section; Authentication using the first audio signal based on the length of the second audio signal of the second utterance interval or the number of sounds included in the second utterance interval by comparing the second audio signal of the interval. An authentication system is provided, which includes a determination unit that determines conditions, and an authentication unit that authenticates the speaker based on the determined authentication conditions.

The present disclosure also provides an authentication method performed by one or more computers, which acquires an audio signal of a speaker's utterance, and determines a first utterance period in which the speaker is speaking based on the acquired audio signal. and a second utterance section in which the speaker is speaking from the voice signals of a database in which the voice signals of each of a plurality of speakers are registered, and the first voice of the first utterance section is detected. The signal is compared with the second audio signal of the second utterance section, and the second utterance is determined based on the length of the second audio signal of the second utterance section or the number of sounds included in the second utterance section. An authentication method is provided, in which authentication conditions for authentication using one audio signal are determined, and the speaker is authenticated based on the determined authentication conditions.

Note that these comprehensive or specific aspects may be realized by a system, an apparatus, a method, an integrated circuit, a computer program, or a recording medium. It may be realized by any combination of the following.

According to the present disclosure, the utterance time during authentication can be determined according to the total length of the user's uttered voice acquired at the time of registration, thereby improving convenience for the user.

A diagram showing an example of a use case of the authentication system according to this embodiment A block diagram showing an example of the internal configuration of an authentication analysis device according to this embodiment Flowchart related to registration processing of uttered audio signal for registration Diagram showing an example of setting authentication conditions based on utterance length Diagram showing an example of setting authentication conditions based on utterance length and number of sounds A diagram showing an example of setting utterance content as an authentication condition according to the quality of the recorded voice signal Diagram showing an example of an operator performing identity verification authentication based on the authentication text displayed on the screen Diagram showing an example of performing identity verification authentication based on the identity verification text displayed on the user side call terminal Diagram showing an example of resetting the required time for authentication conditions based on the measurement results of sound collection conditions during authentication after setting authentication conditions Diagram showing an example of resetting the threshold value of the authentication condition based on the measurement results of the sound collection condition during authentication after setting the authentication condition Flowchart of processing related to speaker authentication Diagram showing an example of setting restrictions on operations after successful authentication depending on the quality of the registered audio signal Flowchart of processing for setting operation restrictions based on the quality of registered audio signals

Hereinafter, embodiments specifically disclosing an authentication system and an authentication method according to the present disclosure will be described in detail with appropriate reference to the drawings. However, more detailed explanation than necessary may be omitted. For example, detailed explanations of well-known matters and redundant explanations of substantially the same configurations may be omitted. This is to avoid unnecessary redundancy in the following description and to facilitate understanding by those skilled in the art. The accompanying drawings and the following description are provided to enable those skilled in the art to fully understand the present disclosure, and are not intended to limit the subject matter of the claims.

First, with reference to FIG. 1, a use case of the authentication system according to the present embodiment will be described. FIG. 1 is a diagram showing an example of a use case of the authentication system according to the present embodiment. The authentication system 100 acquires a voice signal or voice data of a person to be authenticated using voice (in the example shown in FIG. 1, the user US), and stores the acquired voice signal or voice data in a storage ( In the example shown in FIG. 1, the voice signal or voice data of the speaker registered (stored) in the registered speaker database DB) is compared. Based on the matching result, the authentication system 100 evaluates the degree of similarity between the voice signal or voice data collected from the user US who is the authentication target and the voice signal or voice data registered in the storage, and authenticate the user US based on the degree of

The authentication system 100 according to the first embodiment includes an operator-side telephone terminal OP1 as an example of a sound collection device, an authentication analysis device P1, a registered speaker database DB, and a display DP as an example of an output device. configured. Note that the authentication analysis device P1 and the display DP may be integrally configured. Note that the operator-side call terminal OP1 may be replaced with an automatic voice device, and in this case, the automatic voice device may be configured integrally with the authentication analysis device P1.

Note that the authentication system 100 shown in FIG. 1 is used to authenticate a speaker (user US) in a call center as an example, and authenticates the user using voice data collected from the voice of the user US talking to the operator OP. Perform US authentication. The authentication system 100 shown in FIG. 1 further includes a user-side telephone terminal UP1 and a network NW. It goes without saying that the overall configuration of the authentication system 100 is not limited to the example shown in FIG.

The user-side call terminal UP1 is connected to the operator-side call terminal OP1 via the network NW so that they can communicate wirelessly. Note that the wireless communication referred to here is network communication via a wireless LAN (Local Area Network) such as Wi-Fi (registered trademark).

The user-side call terminal UP1 is configured by, for example, a notebook PC, a tablet terminal, a smartphone, a telephone, or the like. The user-side call terminal UP1 is a sound collection device equipped with a microphone (not shown), and collects the voice uttered by the user US, converts it into an audio signal, and transmits the converted audio signal to the operator via the network NW. It is sent to the side call terminal OP1. Further, the user side telephone terminal UP1 acquires the audio signal of the operator OP's utterance transmitted from the operator side telephone terminal OP1, and outputs it from a speaker (not shown).

The network NW is, for example, an IP (Internet Protocol) network or a telephone network, and connects the user-side call terminal UP1 and the operator-side call terminal OP1 to enable transmission and reception of audio signals. Note that data transmission and reception is performed by wired communication or wireless communication.

The operator side telephone terminal OP1 is connected to the user side telephone terminal UP1 and the authentication analysis device P1 so as to be able to transmit and receive data through wired communication or wireless communication, respectively, and transmits and receives audio signals.

The operator-side call terminal OP1 is configured by, for example, a notebook PC, a tablet terminal, a smartphone, a telephone, or the like. The operator-side telephone terminal OP1 acquires an audio signal based on the voice uttered by the user US transmitted from the user-side telephone terminal UP1 via the network NW, and transmits it to the authentication analysis device P1. Note that, when the operator side call terminal OP1 acquires an audio signal including the acquired user US's utterance voice and the operator OP's utterance voice, the operator side call terminal OP1 determines the sound pressure level and frequency band of the voice signal of the operator side call terminal OP1. An audio signal based on the voice uttered by the user US and an audio signal based on the voice uttered by the operator OP may be separated based on voice parameters such as . After separation, the operator-side telephone terminal OP1 extracts only the voice signal based on the voice uttered by the user US and sends it to the authentication analysis device P1.

Furthermore, the operator-side call terminal OP1 may be communicably connected to each of the plurality of user-side call terminals, and may simultaneously acquire audio signals from each of the plurality of user-side call terminals. Operator side call terminal OP1 transmits the acquired audio signal to authentication analysis device P1. Thereby, the authentication system 100 can simultaneously perform voice authentication processing and voice analysis processing for each of a plurality of users.

Furthermore, the operator-side call terminal OP1 may simultaneously acquire audio signals including the voices uttered by each of a plurality of users. The operator-side telephone terminal OP1 extracts a voice signal for each user from each of the voice signals of a plurality of users acquired via the network NW, and transmits the voice signal for each user to the authentication analysis device P1. In such a case, the operator-side call terminal OP1 may analyze the audio signals of multiple users and separate and extract the audio signals for each user based on audio parameters such as sound pressure level and frequency band. . When the voice signals are collected by an array microphone or the like, the operator-side call terminal OP1 may separate and extract the voice signals for each user based on the arrival direction of the uttered voice. As a result, the authentication system 100 can perform voice authentication processing and voice analysis processing for each of a plurality of users, even for voice signals collected in an environment where a plurality of users speak at the same time, such as a web conference, for example. .

An authentication analysis device P1, which is an example of an authentication device and a computer, is connected to the operator side telephone terminal OP1, the registered speaker database DB, and the display DP so that data can be transmitted and received, respectively. Note that the authentication analysis device P1 may be connected to the operator side telephone terminal OP1, the registered speaker database DB, and the display DP via a network (not shown) to enable wired or wireless communication.

The authentication analysis device P1 acquires the voice signal of the user US transmitted from the operator side call terminal OP1, analyzes the acquired voice signal for each frequency, for example, and extracts the utterance features of the user US. . The authentication analysis device P1 refers to the registered speaker database DB, and compares the extracted utterance features with the utterance features of each of the plurality of users registered in advance in the registered speaker database DB. Perform voice authentication of user US. The authentication analysis device P1 generates an authentication result screen SC including the authentication result of the user US, and transmits it to the display DP for output. It goes without saying that the authentication result screen SC shown in FIG. 1 is just an example and is not limited thereto. The authentication result screen SC shown in FIG. 1 includes, for example, a message "The voice matches Mr. Taro Yamada's voice." which is the authentication result of the user US.

The registered speaker database DB, which is an example of a database, is a so-called storage, and is configured using a storage medium such as a flash memory, an HDD (Hard Disk Drive), or an SSD (Solid State Drive). The registered speaker database DB stores (registers) user information of each of a plurality of users and utterance feature amounts in association with each other. The user information here is information related to users, such as a user name, user ID (Identification), or identification information assigned to each user. Note that the registered speaker database DB may be configured integrally with the authentication analysis device P1.

The display DP is configured using, for example, an LCD (Liquid Crystal Display) or an organic EL (Electroluminescence) display, and displays the authentication result screen SC transmitted from the authentication analysis device P1. Note that the display DP may be configured integrally with the authentication analysis device P1.

In the example shown in FIG. 1, the user-side call terminal UP1 collects the user US's uttered voice COM12 ``This is Taro Yamada'' and the uttered voice COM14 ``This is 123245678'', converts them into audio signals, and makes a call to the operator side. Send to terminal OP1. The operator-side telephone terminal OP1 transmits audio signals based on the user US's uttered voices COM12 and COM14 transmitted from the user-side telephone terminal UP1 to the authentication analysis device P1.

The operator-side call terminal OP1 receives the operator OP's uttered voice COM11 "Please tell me your name," the uttered voice COM13 "Please tell me your membership number," and the user US's uttered voice COM12 and uttered voice COM14. When the collected voice signals are obtained, the voice signals based on the operator OP's voice COM11 and voice COM13 are separated and removed, and the voice signals based on the voice COM12 and COM14 of the user US are separated and removed. Only the authentication information is extracted and sent to the authentication analysis device P1. Thereby, the authentication analysis device P1 can improve user authentication accuracy by using only the voice signal of the person to be authenticated.

Next, an example of the internal configuration of the authentication analysis device according to the present embodiment will be described with reference to FIG. 2. FIG. 2 is a block diagram showing an example of the internal configuration of the authentication analysis device according to the present embodiment. The authentication analysis device P1 is configured to include at least a communication unit 20, a processor 21, and a memory 22.

The communication unit 20 is connected to enable data communication with each of the operator-side telephone terminal OP1 and the registered speaker database DB. The communication unit 20 outputs the audio signal transmitted from the operator-side telephone terminal OP1 to the processor 21.

The processor 21 is, for example, a CPU (Central Processing Unit), a DSP (Digital Signal Processor), a GPU (Graphical Processing Unit), or an FPGA (Field). Using a semiconductor chip mounted with at least one of the electronic devices such as Programmable Gate Array) configured. The processor 21 functions as a controller that controls the overall operation of the authentication analysis device P1, and performs control processing to supervise the operation of each part of the authentication analysis device P1, and input/output of data between each part of the authentication analysis device P1. Performs processing, data calculation processing, and data storage processing.

The processor 21 uses the program and data stored in the ROM (Read Only Memory) 22A of the memory 22 to detect a speech interval detection section 21A, a registration quality determination section 21B, a feature amount extraction section 21C, a comparison target setting section 21D, The respective functions of a similarity calculation section 21E, an authentication condition setting section 21F, an authentication sound collection condition measurement section 21G, and an operation restriction setting section 21H are realized. The processor 21 uses a RAM (Random Access Memory) 22B of the memory 22 during operation, and temporarily stores data or information generated or acquired by the processor 21 and each section in the RAM 22B of the memory 22.

The utterance section detection unit 21A, which is an example of a detection unit, acquires an audio signal of the utterance at the time of authentication (hereinafter referred to as "utterance audio signal"), analyzes the acquired utterance audio signal, and determines whether the user US has uttered the utterance. Detects the utterance interval (hereinafter referred to as the first utterance interval) in which the first utterance interval occurs. The utterance section detection section 21A outputs a utterance audio signal (hereinafter referred to as a first audio signal) corresponding to at least one first utterance section detected from the utterance audio signal to the feature extraction section 21C. Further, the speech section detection section 21A may temporarily store the first audio signal of at least one first speech section in the RAM 22B of the memory 22. Note that when the speech section detection section 21A detects a plurality of first speech sections, the speech section detection section 21A may connect the first audio signals of each detected first speech section and output it to the feature amount extraction section 21C. . Furthermore, the speech section detection unit 21A detects a speech section (hereinafter referred to as a second speech section) of the voice data acquired from the user US when registering in advance a speech signal used for authentication of the user US. The speech section detection section 21A outputs a speech audio signal corresponding to the second speech section (hereinafter referred to as a second audio signal) to the registration quality determination section 21B. Note that when there are multiple second speech sections, the speech section detection section 21A connects the second audio signals of the respective detected second speech sections and outputs the concatenated second audio signals to the registration quality determination section 21B. Good too.

The registration quality determination unit 21B, which is an example of a processing unit, acquires a second speech interval or a second audio signal in which each of a plurality of second utterance intervals are concatenated from the utterance interval detection unit 21A. The registration quality determination unit 21B determines the quality of the acquired second audio signal. Quality refers to the quality of the user's surrounding environment at the time of registration, the accuracy of the user's speech, or both, when registering the second audio signal for each user in the registered speaker database DB prior to actual authentication (at the time of registration). This is an indicator that shows the In this embodiment, authentication conditions (see below) to be imposed on the user during actual authentication are determined based on the quality at the time of registration. The registration quality determination unit 21B determines the quality based on, for example, the length of the utterance of the second audio signal (hereinafter referred to as utterance length) or the number of sounds included in the second audio signal. Note that the elements used by the registration quality determining unit 21B to determine the quality are not limited to the utterance length and the number of sounds, but may also be the number of phonemes or the number of words. The registration quality determination unit 21B outputs information on the determined quality to the feature amount extraction unit 21C or the authentication condition setting unit 21F.

The feature extraction unit 21C, which is an example of a processing unit, analyzes the characteristics of the individual's voice, for example, for each frequency, using the one or more utterance audio signals extracted by the utterance section detection unit 21A, and extracts the utterance feature quantity. Extract. The feature extractor 21C extracts the speech feature of the first audio signal of the first speech section output from the speech section detector 21A. Further, the feature extracting section 21C extracts the speech feature of the second audio signal of the second speech section output from the speech section detecting section 21A. Note that the utterance feature amount of the second audio signal of the second utterance section may be registered in advance in the registered speaker database DB. The feature amount extraction unit 21C associates the extracted utterance feature amount of the first utterance section with the first audio signal from which this utterance feature amount is extracted and outputs it to the similarity calculation unit 21E, or It is output to the setting section 21D or temporarily stored in the RAM 22B of the memory 22. The feature amount extraction unit 21C associates the speech feature amount of the second speech section with the second audio signal from which this speech feature amount is extracted and outputs it to the similarity calculation section 21E. The utterance feature amount and the quality information acquired from the registered quality determination unit 21B are linked and temporarily stored in the RAM 22B of the memory 22.

The feature extraction unit 21C performs voice recognition on the utterance content of the utterance audio signal. The method of voice recognition of the utterance content can be realized by a known technique, for example, it may be realized by phoneme analysis of the uttered voice signal and calculated as linguistic information, or it may be realized by other analysis methods.

The comparison target setting unit 21D, which is an example of a setting unit, acquires data on the user US who is the speaker from the registered speaker database DB. Here, the data of the user US is, for example, personal information such as the date of birth, name, or gender of the user US, or at least one of voice data related to utterances registered in the past by the user US, or a feature amount of the voice data. . In order to set the speaker as the user US, the comparison target setting unit 21D may, for example, identify the speaker as the user US using the extracted feature quantity of the speaker output from the feature quantity extracting unit 21C. , the speaker may be identified as the user US based on the content (for example, name or ID) input by the speaker into the user-side call terminal UP1. The comparison target setting section 21D outputs the acquired data of the user US to the utterance section detection section 21A or the similarity calculation section 21E.

The similarity calculation unit 21E, which is an example of the authentication unit, acquires the utterance feature of the utterance audio signal output from the feature extraction unit 21C. The similarity calculating unit 21E calculates the similarity between the utterance feature amount of the first utterance section and the utterance feature amount of the second utterance section obtained from the feature amount extraction section 21C. The similarity calculation unit 21E identifies the user corresponding to the uttered audio signal (that is, the audio signal transmitted from the user-side call terminal UP1) based on the calculated similarity, and executes the user's identity verification authentication. do.

The authentication condition setting unit 21F, which is an example of a determining unit, sets authentication conditions based on the quality-related information acquired from the registered quality determination unit 21B. The authentication conditions include, for example, the length of the speech made by the user US, the content of the speech, or a threshold value for determination. Note that the authentication conditions are not limited to these.

The authentication sound collection condition measurement unit 21G, which is an example of a measurement unit, measures the sound collection conditions at the time of authentication. The sound collection conditions include, for example, the noise, volume, degree of reverberation, or number of phonemes included in the spoken audio signal collected during authentication. Note that the sound collection conditions are not limited to these. The authentication sound collection condition measuring section 21G outputs the measured sound collection conditions to the authentication condition setting section 21F.

The operation restriction setting unit 21H, which is an example of a setting unit, sets restrictions on the operations that the user US can perform based on the quality of the uttered audio signal in the second utterance section. For example, when the authentication system 100 is installed in an ATM (Automatic Teller Machine), the operation restriction setting unit 21H restricts operations such as remittance or transfer when the quality of the spoken audio signal is poor. Note that examples of machines in which the authentication system 100 is installed are not limited to ATMs.

Based on these, the processor 21 sets the authentication conditions for the user's identity verification based on the quality of the second audio signal determined by the registration quality determination unit 21B. The processor 21 acquires the user's uttered audio signal based on the set authentication conditions. The processor 21 determines whether the speaker is the person himself/herself based on the comparison between the first audio signal of the first utterance section detected by the utterance section detection unit 21A and the second audio signal of the second utterance section. authenticate the person.

The memory 22 includes, for example, a ROM 22A that stores programs that define various processes performed by the processor 21 and data used during the execution of the programs, and a ROM 22A that serves as a work memory used when executing various processes performed by the processor 21. It has at least a RAM 22B. A program that defines various processes to be performed by the processor 21 and data used during execution of the program are written in the ROM 22A. The RAM 22B temporarily stores data or information generated or acquired by the processor 21 (for example, the utterance audio signal or the utterance feature amount corresponding to each utterance audio signal).

The display I/F 23 connects the processor 21 and the display DP to enable data communication, and outputs the authentication result screen SC generated by the similarity calculation unit 21E of the processor 21 to the display DP. The display I/F 23 causes the display DP to display an authentication status indicating whether or not the speaker is the authentic speaker based on the authentication result of the processor 21.

Next, with reference to FIG. 3, the registration process of the uttered audio signal for registration will be described. FIG. 3 is a flowchart related to the registration process of the uttered audio signal for registration. Note that each process related to the flowchart in FIG. 3 is executed by the processor 21.

The flowchart in FIG. 3 represents the process related to registration, that is, the registration of the uttered audio signal that is stored in the registered speaker database DB in advance.

The processor 21 starts receiving an utterance audio signal for registration (hereinafter referred to as a registration audio signal) from the speaker (St10). That is, in the process of step St10, the speaker starts speaking to the user-side telephone terminal UP1.

The processor 21 ends receiving the registered audio signal from the speaker (St11). That is, in the process of step St11, the speaker finishes speaking to the user-side telephone terminal UP1.

The speech section detection unit 21A detects the second speech section of the registered audio signal acquired in the processing from step St10 to step St11 (St12).

The registration quality determination unit 21B determines the quality of the second audio signal of the second speech section detected in the process of step St12 (St13).

The registered quality determination unit 21B determines whether or not to re-acquire the registered audio signal based on the quality determined in the process of step St13 (St14). For example, the registration quality determination unit 21B determines not to reacquire when the quality is equal to or higher than a predetermined minimum necessary value, and determines to not reacquire when the quality is less than a predetermined minimum necessary value. judge. For example, if the speaker has not uttered a single word, the length of the utterance is 1 second, or the number of sounds is 1, the registration quality determining unit 21B determines to re-acquire the registered audio signal. Note that the example in which the registration quality determining unit 21B determines to re-acquire is an example and is not limited to these. Moreover, the process of step St14 may be omitted from the process of the flowchart related to FIG. 3.

When the registration quality determination unit 21B determines to re-acquire the registered audio signal (St14, YES), the process of the processor 21 returns to the process of step St10.

When the registration quality determining unit 21B determines that the registered audio signal is not to be reacquired (St14, NO), the feature extracting unit 21C extracts the utterance feature of the uttered audio signal in the second utterance section (St15). .

The feature extracting unit 21C associates the quality determined in the process of step St13 with the utterance feature extracted in the process of step St15 and stores it in the registered speaker database DB (St16).

Next, an example of setting authentication conditions based on utterance length will be described with reference to FIG. 4. FIG. 4 is a diagram illustrating an example of setting authentication conditions based on utterance length.

The registered voice signal US10, the registered voice signal US11, and the registered voice signal US12 are registered voice signals of the user US registered in the registered speaker database DB in the process related to FIG. 3.

In the example shown in FIG. 4, when the utterance length is less than 10 seconds, the quality is "low", and when the utterance length is 10 seconds or more, the quality is "high". Note that the threshold number of seconds for which the quality is "low" or "high" is an example and is not limited. Furthermore, the quality is not limited to two levels of "low" and "high", but may be set to, for example, three levels of "low", "medium" and "high", or four or more levels.

The authentication condition setting unit 21F changes the required time based on the quality results. In the example shown in FIG. 4, when the quality is "low", the required time is 15 seconds, and when the quality is "high", the required time is 7 seconds. The required time is the total speaking time that the authentication system 100 requests from the speaker when performing authentication. Note that the length of the requested time is just an example and is not limited thereto. In the example shown in FIG. 4, the determination threshold is all set to 70 regardless of the quality result. The determination threshold value is a threshold value used by the similarity calculation unit 21E to determine the degree of similarity between the utterance feature amount of the first utterance section and the utterance feature amount of the second utterance section. The higher the determination threshold, the higher the degree of similarity required. Note that the value of the determination threshold is an example and is not limited to 70.

The utterance content of the registered voice signal US10 is "a ka sa tana de su" and the utterance length is 5 seconds. The registered audio signal US10 has a utterance length of 5 seconds and is less than 10 seconds, so the quality is "low". As a result, the required time is 15 seconds and the determination threshold is 70 as authentication conditions for the registered audio signal US10.

The utterance content of the registered audio signal US11 is ``Aka satana de wa ma ya ya ra wa desu'' (a ka sa ta na de su ha ma ya ra wa de su)” and the length of the utterance is 8 seconds. Become. The quality of the registered audio signal US11 is "low" because the utterance length is 8 seconds and it is less than 10 seconds. As a result, the required time is 15 seconds and the determination threshold is 70 as authentication conditions for the registered audio signal US11.

The utterance content of the registered audio signal US12 is ``A ka sa ta na de.'' su　i　chi　ni　sa　n　shi　go　ro　ku　na　na　de su)'' and the length of the utterance is 13 seconds. The quality of the registered audio signal US12 is "high" because the utterance length is 13 seconds and is longer than 10 seconds. As a result, the required time is 7 seconds and the determination threshold is 70 as the authentication conditions for the registered audio signal US12.

Next, an example of setting authentication conditions based on utterance length and number of sounds will be described with reference to FIG. 5. FIG. 5 is a diagram illustrating an example of setting authentication conditions based on utterance length and number of sounds.

In the example shown in FIG. 5, when the utterance length is less than 10 seconds, the quality is "low", and when the utterance length is 10 seconds or more, the quality is "high". When the number of tones is less than 13, the quality is "low", and when the number of tones is 13 or more, the quality is "high". Note that the threshold number of seconds and the number of sounds for which the quality is "low" or "high" are examples and are not limited. Further, the quality is not limited to "low" and "high", but may be set to three stages of "low", "medium" and "high", or to four or more stages.

Based on the quality of the utterance length and number of sounds, the authentication condition setting unit 21F changes the required time. The quality of the registered audio signal is determined by the quality of the utterance length and the quality of the number of sounds, whichever is lower. In the example shown in FIG. 5, when the quality is "low", the required time is 15 seconds, and when the quality is "high", the required time is 7 seconds. Note that the length of the requested time is just an example and is not limited thereto. In the example shown in FIG. 5, the determination threshold is all set to 70 regardless of the quality result.

The utterance content of the registered voice signal US10 is "a ka sa tana de su", and the number of tones is seven. The utterance length of the registered audio signal US10 is 5 seconds, which is less than 10 seconds, so the quality related to the utterance length is "low." Since the number of tones is 7, which is less than 13, the quality related to the number of tones is "low." The quality of the utterance length and the number of tones are both "low", and the quality of the registered audio signal US10 is "low". As a result, the required time as an authentication condition for the registered audio signal US10 is 15 seconds, and the determination threshold is 70.

The utterance content of registered audio signal US11 is "Aka sa ta na de ha ma ya ra wa desu" (a ka sa ta na de su ha ma ya ra) wa de su)" and the number of sounds is 14. . The utterance length of the registered audio signal US11 is 8 seconds, which is less than 10 seconds, so the quality related to the utterance length is "low." The number of tones is 14, which is 13 or more, so the quality related to the number of tones is "high." Since the quality of the number of tones is "high" but the quality of the utterance length is "low", the quality of the registered audio signal US11 is "low". As a result, the required time as an authentication condition for the registered audio signal US11 is 15 seconds, and the determination threshold is 70.

The utterance content of the registered audio signal US12 is ``Aka sa ta na de s. u i chi ni sa n shi go ro ku na na de su )” and the number of tones is 20. The utterance length of the registered audio signal US12 is 13 seconds, which is 10 seconds or more, so the quality related to the utterance length is "high." Since the number of tones is 20, which is 13 or more, the quality related to the number of tones is "high." The quality of the utterance length and the number of tones are both "high", and the quality of the registered audio signal US12 is "high". As a result, the required time as an authentication condition for the registered audio signal US12 is 7 seconds, and the determination threshold is 70.

Next, with reference to FIG. 6, an example of setting the utterance content as an authentication condition according to the quality of the registered audio signal will be described. FIG. 6 is a diagram illustrating an example of setting utterance content as an authentication condition according to the quality of a registered audio signal. The quality determination method in FIG. 6 is the same as the determination method in FIG. 4.

The authentication condition setting unit 21F specifies words that prompt the user US to speak based on the quality results. In the example shown in FIG. 6, when the quality is "low", the feature quantity extracting unit 21C performs voice recognition of the registered voice signal, analyzes the utterance content, and outputs it to the authentication condition setting unit 21F. The authentication condition setting unit 21F determines the wording to be specified for the user US based on the voice recognition result obtained from the feature extracting unit 21C. When the quality is "low" and the authentication condition setting unit 21F specifies the utterance content, the requested time is not specified. If the quality is "high", no text is specified for the user US. Note that even if the quality is "high", the user US may be allowed to specify a text, or a shorter text may be specified than when the quality is "low" to improve user convenience. In the example shown in FIG. 6, when the quality is "high", the required time is set to 7 seconds as the authentication condition. Note that the required time when the quality is "high" is an example and is not limited to 7 seconds. In the example of FIG. 6, the determination threshold is set to a constant value of 70 regardless of quality. Note that the determination threshold value may not be a constant value but may be changed depending on the quality.

The utterance content of the registered voice signal US10 is "Akasatana desu". Since the quality of the registered audio signal US10 is "low", the feature extraction unit 21C performs speech recognition of the registered audio signal US10, and the recognition result is "Akasatana Desu". The authentication condition setting unit 21F specifies the wording as “Akasatana desu” based on the recognition result obtained from the feature extracting unit 21C. The authentication condition setting unit 21F sets the request time to no specification and the determination threshold to 70.

The utterance content of the registered audio signal US12 is "Akasatana desu, Ichinisan Shigorokunana desu." Since the quality of the registered audio signal US12 is "high", the feature extraction unit 21C does not perform speech recognition. The authentication condition setting unit 21F sets the required time to 7 seconds and the determination threshold to 70 as authentication conditions. The authentication condition setting unit 21F does not specify the wording because the quality of the registered audio signal US12 is "high".

As a result, the authentication system 100 can authenticate in a short time while maintaining high authentication accuracy by specifying and having the user US utter a phrase based on the utterance content of the registered voice signal according to the quality of the registered voice signal. . Furthermore, when the quality of the registered voice signal is high, the authentication system 100 does not specify the content of the utterance, thereby saving the user US's effort.

Next, with reference to FIG. 7, an example will be described in which the operator performs identity verification authentication based on the identity verification text displayed on the screen. FIG. 7 is a diagram illustrating an example in which an operator performs identity verification based on the authentication text displayed on the screen.

When the quality shown in FIG. 6 is "low", the authentication system 100 uses voice recognition to specify the words to be uttered by the user US. In FIG. 7, the operator OP displays the message to be uttered by the user US designated by the method shown in FIG. An example of making US utter a specified phrase will be explained.

First, an example shown in case CC will be explained. Screen SC1 is an example of an operator screen displayed on display DP.

Speaker registration information is displayed in frame FR1. In frame FR1, "sender number", "registered name", "registered address", "age", and "speaker registration status" are displayed as speaker registration information. The "source number" is, for example, a telephone number. “Speaker registration presence/absence” indicates whether or not a registered voice signal is stored in the registered speaker database DB. Furthermore, if the registered voice signal is stored in the registered speaker database DB, the quality associated with the registered voice signal is also displayed. For example, in frame FR1, the "sender number" is XXX-XXXXXX-XXXXXX, the "registered name" is Ada A, the "registered address" is ABCDEFG, the "age" is 33, and " ``Speaker registration presence/absence'' is displayed as Yes (quality: low).

In frame FR2, candidates who are considered to be speakers are displayed as the speaker authentication results. Furthermore, the probability that the speaker is a candidate is displayed next to the candidate. In the example of frame FR2, the probability is expressed as a percentage, but the probability is not limited to this and may be expressed as "low, medium, high." In frame FR2, "A-field A-man: 70%", "B-yama B-ro: 25%" and "C-kawa C-husband: 5%" are displayed as the authentication results.

In the frame FR3, a sentence to be uttered by the speaker (hereinafter referred to as an identity authentication sentence) is displayed. In frame FR3, "I am Hamayarawa, I am Hachikyujuzero" is displayed as a text for personal authentication.

The button BT1 is a button to start or stop authentication.

Operator OP utters utterance voice OP10, "I'm Hamayarawa. Please say, 'I'm Hachikyujuzero.'" based on the text for personal authentication displayed in frame FR3 of screen SC1. Based on the utterance of the operator OP, the user US utters the utterance voice US13, "It's Hamayarawa. It's Hachikyujuzero."

Next, the example shown in case CD will be explained. Screen SC2 is an example of an operator screen displayed on display DP.

Speaker registration information is displayed in frame FR5. In the frame FR5, "caller number", "registered name", "registered address", "age", and "speaker registration status" are displayed as speaker registration information. For example, in frame FR2, the "sender number" is XXX-XXXXXX-XXXXXX, the "registered name" is B-Yama Brou, the "registered address" is GFEDCBA, the "age" is 44, and " ``Speaker registration presence/absence'' is displayed as Yes (quality: high).

In the frame FR6, candidates who are considered to be speakers are displayed as the speaker authentication results. Furthermore, the probability that the speaker is a candidate is displayed next to the candidate. In frame FR2, "A-field A-man: 15%", "B-yama B-ro: 60%", and "C-kawa C-husband: 25%" are displayed as the authentication results.

In frame FR7, "Hamayarawa Desu" is displayed as the text for personal authentication. Since the quality of the case CD is "high", a sentence shorter than that of the case CC whose quality is "low" is designated as the identity verification sentence.

The button BT2 is a button to start or stop authentication.

Operator OP utters utterance voice OP11, "Please say, 'It's Hamayarawa.'" based on the personal authentication text displayed in frame FR7 of screen SC2. Based on the utterance of the operator OP, the user US utters the uttered voice US14, "It's Hamaya Rawa."

In FIG. 7, the operator OP reads out the personal authentication text displayed on the operator screen and has the user US speak it, but the user US may also read the personal authentication text using an automatic voice and have the user US speak it.

Thereby, the authentication system 100 allows the user US and the operator OP to perform authentication without worrying about the length of the utterance. Further, when the quality of the registered audio signal is high, the authentication system 100 can specify a short phrase to the user US for authentication, thereby reducing the time required for authentication. Furthermore, when the quality of the registered audio signal is low, the authentication system 100 can maintain high authentication accuracy by specifying a longer word for the user US than when the quality is high, thereby preventing authentication failure or redoing. be able to.

Next, with reference to FIG. 8, an example will be described in which identity verification is performed based on the identity verification text displayed on the user side call terminal. FIG. 8 is a diagram illustrating an example of performing identity verification authentication based on the identity verification text displayed on the user side call terminal.

First, an example shown in case CE will be explained. Case CE is an example where the quality of the registered audio signal is low. Screen SC3 is an example of a screen displayed on user side call terminal UP1.

On the screen SC3, the following message is displayed: "Please speak the following text to confirm your identity: This is Hamayarawa. This is Hachikyujuzero." In frame FR9, "It's Hamayarawa, I'm Hachikyujuzero" is displayed as the identity verification text, and the text displayed differs depending on the user US.

The user US looks at the content displayed on the screen SC3 and utters the utterance voice US13, "It's Hamayarawa. It's Hachikyujuzero."

Next, an example shown in case CF will be explained. Case CF is an example where the quality of the registered audio signal is high. Screen SC4 is an example of a screen displayed on user side call terminal UP1.

On the screen SC4, the message "Please speak the identity verification sentence: Hamayarawa desu" is displayed. In frame FR9, "Hamayarawa desu" is displayed as the identity verification text.

The user US looks at the content displayed on the screen SC4 and utters the utterance voice US13, "It's Hamayarawa."

Thereby, the authentication system 100 can authenticate the user US without worrying about the length of his or her utterance. Furthermore, with this, the authentication system 100 can authenticate the user US unattended without the intervention of a person such as the operator OP.

Next, with reference to FIGS. 9 and 10, an example will be described in which the authentication conditions are reset based on the measurement results of the sound collection conditions during authentication after the authentication conditions are set. FIG. 9 is a diagram illustrating an example of resetting the required time of the authentication condition based on the measurement result of the sound collection condition at the time of authentication after setting the authentication condition. FIG. 10 is a diagram showing an example of resetting the threshold value of the authentication condition based on the measurement result of the sound collection condition at the time of authentication after setting the authentication condition.

The authentication sound collection condition measuring unit 21G determines the noise, volume, degree of reverberation, or speech sound signal of the speech sound signal collected during authentication as the sound collection condition during authentication (hereinafter referred to as the sound collection condition during authentication). Measure the number of phonemes included. FIGS. 9 and 10 show an example in which the authentication conditions are once set (hereinafter referred to as initial authentication conditions) and then are reset based on the measured authentication sound collection conditions.

If the noise of the uttered audio signal is greater than a predetermined value regarding noise, that is, if there is a lot of noise, the authentication condition setting unit 21F increases the required time by 3 seconds as an authentication condition.

If the volume of the uttered audio signal is less than the predetermined value regarding the volume, that is, if the volume is low, the authentication condition setting unit 21F increases the required time by 3 seconds as the authentication condition.

If the number of phonemes in the uttered audio signal is less than a predetermined value regarding the number of phonemes, that is, if the number of phonemes is small, the authentication condition setting unit 21F increases the required time by 3 seconds as an authentication condition.

When the reverberation of the uttered audio signal is greater than or equal to a predetermined value regarding reverberation, that is, when the reverberation is large, the authentication condition setting unit 21F increases the required time by 5 seconds as an authentication condition.

Note that the length of the required time to be increased regarding noise, volume, number of phonemes, and reverberation is an example and is not limited to these.

When the quality of the spoken audio signal is "low", the initial authentication conditions are a required time of 15 seconds and a determination threshold of 70. Note that the initial authentication conditions are just an example and are not limited thereto. If there is a lot of noise as the sound collection condition for authentication, the authentication condition setting unit 21F increases the required time by 3 seconds. As a result, the authentication conditions after resetting are a request time of 18 seconds and a determination threshold of 70. If the volume is low as the sound collection condition for authentication, the authentication condition setting unit 21F increases the required time by 3 seconds. As a result, the authentication conditions after resetting are a request time of 18 seconds and a determination threshold of 70.

When the quality of the uttered audio signal is "high", the initial authentication conditions are a required time of 7 seconds and a determination threshold of 70. Note that the initial authentication conditions are just an example and are not limited thereto. When the sound collection conditions for authentication are low volume and a lot of noise, the authentication condition setting unit 21F increases the required time by 6 seconds in total. As a result, the authentication conditions after resetting are a request time of 13 seconds and a determination threshold of 70. When the number of phonemes is small as the sound collection condition for authentication, the authentication condition setting unit 21F increases the required time by 3 seconds. As a result, the authentication conditions after resetting are a request time of 10 seconds and a determination threshold of 70. When the reverberation is large as the sound collection condition for authentication, the authentication condition setting unit 21F increases the required time by 5 seconds. As a result, the authentication conditions after resetting are a request time of 12 seconds and a determination threshold of 70. If the sound collection conditions at the time of authentication are good, the authentication conditions are the same as the initial authentication conditions.

Next, with reference to FIG. 10, an example will be described in which the threshold value of the authentication condition is reset based on the measurement result of the sound collection condition at the time of authentication after the authentication condition is set.

If the noise in the uttered audio signal is greater than or equal to a predetermined noise-related value, that is, if there is a lot of noise, the authentication condition setting unit 21F lowers the determination threshold by 10 as the authentication condition.

If the volume of the uttered audio signal is less than the predetermined value regarding the volume, that is, if the volume is small, the authentication condition setting unit 21F lowers the determination threshold by 15 as the authentication condition.

When the number of phonemes in the uttered audio signal is less than a predetermined value regarding the number of phonemes, that is, when the number of phonemes is small, the authentication condition setting unit 21F lowers the determination threshold by 10 as an authentication condition.

When the reverberation of the uttered audio signal is greater than or equal to a predetermined value regarding reverberation, that is, when the reverberation is large, the authentication condition setting unit 21F lowers the determination threshold by 20 as the authentication condition.

Note that the values of the determination thresholds to be lowered with respect to noise, volume, number of phonemes, and reverberation are only examples, and are not limited to these.

When the quality of the spoken audio signal is "low", the initial authentication conditions are a required time of 15 seconds and a determination threshold of 70. Note that the initial authentication conditions are just an example and are not limited thereto. If there is a lot of noise as the sound collection condition for authentication, the authentication condition setting unit 21F lowers the determination threshold by 10. As a result, the authentication conditions after resetting are a request time of 15 seconds and a determination threshold of 60. When the volume is low as the sound collection condition at the time of authentication, the authentication condition setting unit 21F lowers the determination threshold value by 15. As a result, the authentication conditions after resetting are a request time of 15 seconds and a determination threshold of 55.

When the quality of the uttered audio signal is "high", the initial authentication conditions are a required time of 7 seconds and a determination threshold of 70. Note that the initial authentication conditions are just an example and are not limited thereto. When the sound collection conditions at the time of authentication are that the volume is low and there is a lot of noise, the authentication condition setting unit 21F lowers the determination threshold by 25 in total. As a result, the authentication conditions after resetting are a request time of 7 seconds and a determination threshold of 45. When the number of phonemes is small as the sound collection condition for authentication, the authentication condition setting unit 21F lowers the determination threshold by 10. As a result, the authentication conditions after resetting are a request time of 7 seconds and a determination threshold of 60. When the reverberation is large as a sound collection condition at the time of authentication, the authentication condition setting unit 21F lowers the determination threshold value by 20. As a result, the authentication conditions after resetting are a request time of 7 seconds and a determination threshold of 50. If the sound collection conditions at the time of authentication are good, the authentication conditions are the same as the initial authentication conditions.

Thereby, regardless of the quality of the registered audio signal, the authentication system 100 can perform authentication with high accuracy by lengthening the request time, lowering the determination threshold, or both if the sound collection conditions at the time of authentication are poor.

Next, processing related to speaker authentication will be described with reference to FIG. 11. FIG. 11 is a flowchart of processing related to speaker authentication. Each process related to FIG. 11 is executed by the processor 21.

The comparison target setting unit 21D sets a person to be used for authentication from among a plurality of people registered in the registered speaker database DB when performing authentication to confirm the identity of a speaker who is an authentication target person (St20). .

The comparison target setting unit 21D acquires information regarding the quality of the registered voice signal of the person who is the comparison target set in the process of step St20 from the registered speaker database DB (St21). The comparison target setting section 21D outputs the acquired information to the authentication condition setting section 21F. Further, the comparison target setting unit 21D acquires the registered feature amount of the registered voice signal of the person who is the comparison target from the registered speaker database DB, and outputs it to the similarity calculation unit 21E.

The authentication sound collection condition measurement unit 21G measures the authentication sound collection conditions (St22). Note that the process of step St22 may be omitted from the process of the flowchart related to FIG. 11.

The authentication condition setting unit 21F sets authentication conditions based on the information regarding quality acquired from the comparison target setting unit 21D in the process of step St21 (St23).

The processor 21 transmits a signal to start the authentication process to the communication unit 20 (St24). The communication unit 20 transmits an instruction to the operator side telephone terminal OP1 to start authentication.

The authentication condition setting unit 21F acquires the registered feature amount of the speaker's registered voice signal from the comparison target setting unit 21D. The authentication condition setting unit 21F specifies the wording of the utterance content to be used for authentication based on the acquired registered feature amount (St25). Note that the process of step St25 may be omitted from the process of the flowchart related to FIG. 11.

The processor 21 starts receiving the uttered audio signal used for authentication (St26). The processor 21 outputs the acquired speech audio signal to the feature extraction unit 21C.

The authentication sound collection condition measurement unit 21G measures the authentication sound collection conditions (St27). The authentication sound collection condition measuring unit 21G outputs information on the measured authentication sound collection conditions to the authentication condition setting unit 21F. Note that the process of step St27 may be omitted from the process of the flowchart related to FIG. 11.

The authentication condition setting unit 21F resets the authentication condition based on the authentication sound collection condition acquired in the process of step St27 (St28). The processor 21 acquires the uttered audio signal based on the authentication condition reset by the authentication condition setting unit 21F in the process of step St28. The processor 21 outputs the acquired speech audio signal to the feature extraction unit 21C. Note that the process of step St28 may be omitted from the process of the flowchart related to FIG. 11.

The processor 21 transmits a signal to the communication unit 20 to end the authentication process, that is, to end the reception of the uttered audio signal used for authentication (St29). The communication unit 20 transmits an instruction to end the authentication to the operator side call terminal OP1.

The feature amount extraction unit 21C extracts the speech feature amount of the speech audio signal acquired in the process of step St26 or the process of step St28 (St30). The feature extraction unit 21C outputs the extracted utterance feature to the similarity calculation unit 21E.

The similarity calculation unit 21E calculates the similarity based on the registered feature obtained in step St21 and the utterance feature obtained in the process of step St30 (St31).

The similarity calculation unit 21E determines whether the similarity calculated in the process of step St31 is greater than or equal to a predetermined threshold (St32). When the similarity calculation unit 21E determines that the similarity is greater than or equal to a predetermined threshold (St32, YES), the similarity calculation unit 21E sends a signal to the communication unit 20, display I/F 23, or Output to both.

If the similarity calculation unit 21E determines that the similarity is less than a predetermined threshold (St32, NO), the similarity calculation unit 21E determines whether or not to continue the authentication process (St34).

When the similarity calculation unit 21E determines to continue the authentication process (St34, YES), the process of the processor 21 returns to the process of step St22. If the process of step St22 is omitted, the process of the processor 21 returns to the process of step St23.

If the similarity calculation unit 21E determines not to continue the authentication process (St35, NO), it outputs a signal to the communication unit 20, the display I/F 23, or both to the effect that the authentication of the speaker's identity verification has failed. .

Next, with reference to FIG. 12, an example will be described in which restrictions are placed on operations after successful authentication depending on the quality of the registered audio signal. FIG. 12 is a diagram illustrating an example in which restrictions are placed on operations after successful authentication depending on the quality of the registered audio signal.

Registered audio signal US10 and registered audio signal US12 according to FIG. 12 are similar to registered audio signal US10 and registered audio signal US12 according to FIG. 4. Therefore, in FIG. 12, the explanation up to the setting of the authentication conditions is omitted.

When the authentication system 100 is installed in a bank ATM, for example, the operation restriction setting unit 21H sets the operation (for example, deposit) after the speaker's identity verification has been successfully authenticated based on the quality of the registered audio signal. You can put restrictions on it. Although examples in which the authentication system 100 is installed are not limited to bank ATMs, here, for convenience of explanation, it is assumed that the authentication system 100 is installed in a bank ATM.

Case CG is an example of authentication when the quality of the registered audio signal is low. In case CG, since the quality of the registered audio signal is low, the operation restriction setting unit 21H restricts the operation mode and operates in the restriction mode. In the restricted mode, for example, only the introduction and deposit of the balance of the account can be made. Note that the operations that are possible in the restricted mode are merely examples and are not limited to these.

Case CH is an example of authentication when the quality of the registered audio signal is high. In case CH, since the quality of the registered audio signal is high, the operation restriction setting unit 21H operates in a normal mode in which no restrictions are placed on the operation mode. In the normal mode, all operations such as introduction of remaining balance in the account, deposit, remittance, or transfer are possible. Note that the operations that are possible in the normal mode are just examples and are not limited to these.

Thereby, when the quality of the registered audio signal is low, the authentication system 100 can reduce the risk of misjudgment by placing operational restrictions on the machine in which the authentication system 100 is installed.

Next, with reference to FIG. 13, a process for setting operation restrictions based on the quality of registered audio signals will be described. FIG. 13 is a flowchart of a process for setting operation restrictions based on the quality of registered audio signals. Each process in the flowchart in FIG. 13 is executed by the processor 21. It should be noted that processes in the flowchart of FIG. 13 that are similar to those in the flowchart of FIG. 11 are given the same reference numerals, and description thereof will be omitted.

If the authentication related to the speaker's identity verification is successful in the process of step St34, the operation restriction setting unit 21H determines whether the quality of the registered audio signal is high (St36).

When the operation restriction setting unit 21H determines that the quality of the registered audio signal is high (St36, YES), it sets the operation mode to the normal mode (St37).

If the operation restriction setting unit 21H determines that the quality of the registered audio signal is low (St36, NO), it sets the operation mode to the restriction mode (St38).

As described above, the authentication system (eg, authentication system 100) according to the present embodiment includes an acquisition unit (eg, user-side call terminal UP1) that acquires the audio signal of the speaker's uttered voice. The authentication system determines the first utterance period in which the speaker is speaking based on the acquired audio signal, and the first utterance period in which the speaker is speaking based on the audio signal of the database in which the audio signals of each of a plurality of speakers are registered. 2 (for example, the speech section detection section 21A). The authentication system compares the first audio signal of the first utterance section with the second audio signal of the second utterance section, and determines the length of the second audio signal of the second utterance section or the second speech section. includes a determining unit (for example, an authentication condition setting unit 21F) that determines authentication conditions for authentication using the first audio signal based on the number of sounds included in the first audio signal. The authentication system includes an authentication section (eg, similarity calculation section 21E) that authenticates the speaker based on the determined authentication conditions.

As a result, the authentication system according to the present embodiment can determine the authentication conditions required of the user during authentication based on the utterance length or number of sounds of the registered audio signal, so the authentication conditions can be changed for each user. . Thereby, the authentication system can determine the utterance time at the time of authentication according to the total length of the user's utterance voice acquired at the time of registration, and improve convenience for the user.

Further, the authentication unit of the authentication system according to the present embodiment is configured such that the length of the second utterance section is equal to or greater than the first predetermined value, and the length of the first utterance section is equal to or greater than the second predetermined value. Authentication will start if the The authentication unit performs authentication when the length of the second utterance section is less than a first predetermined value and the length of the first utterance section is greater than or equal to a third predetermined value that is larger than the second predetermined value. Start. As a result, the authentication system requests the user to speak by setting a number of seconds that is estimated to be sufficient for authentication at the time of authentication, based on the quality of the registered audio signal based on the length of the utterance. This can prevent authentication from failing due to too short utterances. Thereby, the authentication system 100 can determine the utterance time at the time of authentication according to the length of the total time of the user's uttered voice acquired at the time of registration, and can improve convenience for the user.

Further, the authentication unit of the authentication system according to the present embodiment is configured such that the number of sounds included in the second speech section is equal to or greater than a fourth predetermined value, and the length of the first speech section is equal to or greater than a second predetermined value. Authentication starts when the above conditions are met. When the number of sounds included in the second utterance section is less than a fourth predetermined value and the length of the first utterance section is greater than or equal to a third predetermined value that is greater than the second predetermined value, the authentication section Start authentication. As a result, the authentication system requests the user to speak by setting a number of seconds that is estimated to be sufficient for stability during authentication based on the quality of the registered audio signal based on the number of tones. This can prevent authentication from failing due to too short utterances. Thereby, the authentication system can improve user convenience during authentication.

Further, the authentication unit of the authentication system according to the present embodiment is configured such that the length of the second speech section is equal to or greater than the first predetermined value, and the number of sounds included in the second speech section is equal to or greater than the fourth predetermined value. Then, the authentication is started when the length of the first utterance section is equal to or greater than the second predetermined value. The authentication unit is configured to determine whether the length of the second speech section is less than a first predetermined value or the number of sounds included in the second speech section is less than a fourth predetermined value, and the length of the first speech section is less than a first predetermined value. Authentication is started when the third predetermined value is greater than the second predetermined value. Thereby, the authentication system can determine the utterance length required of the user at the time of authentication, according to the utterance length and the number of sounds of the registered audio signal. Since the authentication system can determine authentication conditions based on the length of the utterance and the number of sounds, it can improve user convenience and perform more accurate authentication.

Further, when the length of the second speech section is less than the first predetermined value, the determination unit of the authentication system according to the present embodiment determines the length of the speech section from the speech content included in the audio signal of the second speech section. Decide on the text that prompts you to speak. Thereby, the authentication system can perform authentication in a short time while maintaining high authentication accuracy by having the user specify and utter a phrase based on the utterance content of the registered voice signal according to the utterance length of the registered voice signal. Furthermore, if the utterance length of the registered audio signal is sufficiently long, the authentication system does not specify the content of the utterance, thereby saving the user US effort.

Further, the authentication system according to the present embodiment further includes a first display unit (for example, display DP) that displays a screen that the operator refers to during authentication, and the determination unit causes the first display unit to display text. Thereby, the authentication system allows the user and the operator to perform authentication without worrying about the length of the utterance. Furthermore, if the quality of the registered audio signal is high, the authentication system can specify a short phrase to the user for authentication, thereby reducing the time required for authentication. In addition, when the quality of the registered audio signal is low, the authentication system can maintain high authentication accuracy by specifying a longer word for the user than when the quality is high, and prevent authentication failure or retry. can.

Further, the authentication system according to the present embodiment further includes a second display unit (for example, user-side call terminal UP1) that displays a screen that the speaker refers to during authentication, and the determination unit displays the text on the second display unit. Display. The authentication system can perform authentication without requiring the user to worry about the length of the utterance. Furthermore, this allows the authentication system to authenticate users unattended without the intervention of a person such as an operator.

Further, the authentication system according to the present embodiment further includes a measurement unit that measures at least one of noise, volume, number of phonemes, or reverberation size of the audio signal in the first speech section, and the determination unit measures Authentication conditions are set based on the measurement results obtained from the department. As a result, regardless of the quality of the registered audio signal, the authentication system can perform authentication with high accuracy by lengthening the required time, lowering the determination threshold, or both if the sound collection conditions at the time of authentication are poor.

Further, the authentication condition of the authentication system according to the present embodiment is the length of the audio signal or the threshold value for determination related to the authentication of the speaker's identity. Thereby, the authentication system can specify the utterance length to be specified to the user according to the quality of the registered audio signal of each user, and can improve convenience for the user. Further, the authentication system can change the threshold value for determination depending on the quality of the user's registered audio signal, and can perform flexible authentication depending on each user.

In addition, the authentication system according to the present embodiment is configured such that when the length of the second utterance section is less than the first predetermined value, a restriction setting unit (for example, an operation restriction It further includes a setting section 21H). Thereby, when the quality of the registered audio signal is low, the authentication system can lower the risk of misjudgment by placing operational restrictions on the machine in which the authentication system is installed.

Although the embodiments have been described above with reference to the accompanying drawings, the present disclosure is not limited to such examples. It is clear that those skilled in the art can come up with various changes, modifications, substitutions, additions, deletions, and equivalents within the scope of the claims, and It is understood that it falls within the technical scope of the present disclosure. Further, each of the constituent elements in the embodiments described above may be combined as desired without departing from the spirit of the invention.

Note that this application is based on a Japanese patent application (Japanese Patent Application No. 2022-086892) filed on May 27, 2022, and the contents thereof are incorporated as a reference in this application.

The technology of the present disclosure is useful as an authentication system and method that improves user convenience by determining the utterance time during authentication according to the total length of the user's utterances acquired at the time of registration.

NW network UP1 User-side call terminal OP1 Operator-side call terminal US User OP Operator COM11, COM12, COM13, COM14 Speech audio P1 Authentication analysis device DB Registered speaker database DP Display SC Authentication result screen 20 Communication unit 21 Processor 21A Speech section detection unit 21B Registration quality determination unit 21C Feature extraction unit 21D Comparison target setting unit 21E Similarity calculation unit 21F Authentication condition setting unit 21G Authentication sound collection condition measurement unit 22H Operation restriction setting unit 22 Memory 22A ROM
22B RAM
23 Display I/F
US10, US11, US12 Registered audio signal CA, CB, CC, CD, CE, CF, CG, CH Case SC1, SC2, SC3, SC4 Screen FR1, FR2, FR3, FR4, FR5, FR6, FR7 Frame BT1, BT2 Button OP10, OP11, US13, US14 Utterance audio

Claims

an acquisition unit that acquires an audio signal of the speaker's uttered voice;
The first utterance period in which the speaker is speaking is determined from the acquired audio signal, and the first utterance period in which the speaker is speaking is determined from the audio signal in the database in which the audio signals of each of a plurality of speakers are registered. a detection unit that detects the utterance section of 2;
The first audio signal of the first utterance section and the second audio signal of the second utterance section are compared, and the length of the second audio signal of the second utterance section or the second utterance is determined. a determining unit that determines authentication conditions for authentication using the first audio signal based on the number of sounds included in the section;
an authentication unit that authenticates the speaker based on the determined authentication condition;
Authentication system.
The authentication section is
starting the authentication when the length of the second utterance section is equal to or greater than a first predetermined value, and the length of the first utterance section is equal to or greater than a second predetermined value;
When the length of the second utterance section is less than the first predetermined value and the length of the first utterance section is greater than or equal to the third predetermined value, which is larger than the second predetermined value, start authentication,
The authentication system according to claim 1.
The authentication section is
starting the authentication when the number of sounds included in the second speech section is equal to or greater than a fourth predetermined value, and the length of the first speech section is equal to or greater than a second predetermined value;
The number of sounds included in the second speech section is less than the fourth predetermined value, and the length of the first speech section is greater than or equal to a third predetermined value that is greater than the second predetermined value. initiating said authentication if
The authentication system according to claim 1.
The authentication section is
The length of the second speech section is greater than or equal to a first predetermined value, the number of sounds included in the second speech section is greater than or equal to a fourth predetermined value, and the length of the first speech section is greater than or equal to a fourth predetermined value. Starting the authentication when the value is equal to or higher than a second predetermined value;
The length of the second speech section is less than the first predetermined value, or the number of sounds included in the second speech section is less than the fourth predetermined value, and the length of the first speech section is less than the first predetermined value. starting the authentication when the value is equal to or greater than a third predetermined value that is larger than the second predetermined value;
The authentication system according to claim 1.
The determining unit is
If the length of the second speech section is less than a first predetermined value, determining a text that prompts the speaker to speak from the speech content included in the audio signal of the second speech section;
The authentication system according to claim 1.
further comprising a first display unit that displays a screen that the operator refers to during the authentication;
The determining unit displays the text on the first display unit,
The authentication system according to claim 5.
further comprising a second display unit that displays a screen that the speaker refers to during the authentication;
The determining unit displays the text on the second display unit,
The authentication system according to claim 5.
Further comprising a measurement unit that measures at least one of noise, volume, number of phonemes, or reverberation size of the audio signal in the first speech section,
The determining unit sets the authentication condition based on the measurement result obtained from the measuring unit.
The authentication system according to claim 5.
The authentication condition is a length of the audio signal or a threshold value for determining the identity of the speaker.
The authentication system according to claim 8.
If the length of the second utterance section is less than a first predetermined value, the speaker further includes a limit setting unit that sets a limit on operations that the speaker can perform after authentication.
The authentication system according to claim 1.
An authentication method performed by one or more computers, the method comprising:
Obtain the audio signal of the speaker's utterance,
The first utterance period in which the speaker is speaking is determined from the acquired audio signal, and the first utterance period in which the speaker is speaking is determined from the audio signal in the database in which the audio signals of each of a plurality of speakers are registered. Detecting the utterance section of 2,
collating a first audio signal of the first utterance section and a second audio signal of the second utterance section;
determining authentication conditions for authentication using the first audio signal based on the length of the second audio signal of the second utterance section or the number of sounds included in the second utterance section;
authenticating the speaker based on the determined authentication conditions;
Authentication method.